| With the development of industry and commerce,the serious air pollution has made an adverse impact on people’s lives.Thus,air quality has been more and more concerned.It is becoming research hotspots to analyze the data collected from different monitoring stations,mini useful information from big data,optimize the layout of stations,detect abnormal data,and predict air pollution.The Ministry of Ecology and Environment has deployed monitoring stations(state-control stations)in all districts and ranked air quality mainly based on the data collected from the stations.Meanwhile,local governments have set up micro stations around the state-control stations.To collect the data from micro stations and analyze the contributing rate(i.e.correlation)to the national-control stations of each one of the micro stations at different times and in different regions so as to lower the adverse impact of air pollution on people.The main work is as follows:(1)Analyzed and designed a visualized analysis system of air quality data.The map visualization and chart visualization are achieved with the application of the Gaode map API and ECharts respectively.Ajax asynchronous data transmission interface is used to mark the positions of monitoring stations differently according to latitude and longitude data on Gaode map.The information form is designed to display site specific information.A set of charts is designed to visualize the variation trend of each pollutant concentration.(2)Calculated a correlation between state-control station and micro stations.The Spearman correlation coefficient is used to calculate the correlation between state-control station and micro station in different time and regions regarding to.The contributing rate of each micro station to state-control station is measured according to the magnitude of the correlation.The local governments will curb the air condition and optimize the layout of stations according to the rate.(3)Designed the detection solution for the abnormal pollutant concentration.The data of pollutant concentration is a time sequence.Accordingly,the One-Class SVM,due to its functionality is used to defect abnormal data.(4)Designed and implemented a prediction solution of pollutant concentration.A predictive model based on RBF neural network is employed.Based on historical data,the concentration of each pollutant in the next day is predicted.According to the prediction results,steps are taken ahead to warm early,hence reduce the adverse effects of air pollution on people. |
